1
Reference
Document:
"
Source
apportionment
of
PM2.5
using
a
three­
dimensional
air
quality
model
and
a
receptor
model,"
Park,
S­
K,
L.
Ke,
B.
Yan,
A.
G.
Russell,
M.
Zheng
(
2005),
Proceedings
of
an
AAAR
international
specialty
conference
­­
Particulate
Matter
Supersites
Program
and
Related
Studies,
Atlanta,
Georgia.

Source
apportionment
of
PM
2.5
using
a
three­
dimensional
air
quality
model
and
a
receptor
model
Sun­
Kyoung
Park*,
Lin
Ke**,
Bo
Yan**,
Armistead
G.
Russell*,
Mei
Zheng**

*
School
of
Civil
and
Environmental
Engineering,
Georgia
Institute
of
Technology,
GA
**
School
of
Earth
and
Atmospheric
Sciences,
Georgia
Institute
of
Technology,
GA
Sources
of
PM2.5
are
apportioned
using
a
three­
dimensional
air
quality
model
and
a
receptor
model.
Objectives
of
this
paper
are
to
compare
source
contributions
calculated
from
the
two
modeling
approaches,
and
to
identify
origins
of
discrepancies.
In
addition,
this
paper
investigates
ways
by
which
methods
of
source
apportionment
can
be
improved.
Air
quality
modeling
is
done
with
CMAQ
applied
over
the
United
States,
with
a
detailed
grid
over
the
southeastern
US,
and
receptor
modeling
is
conducted
using
the
chemical
mass
balance
model
with
organic
molecular
markers
(
CMB­
MM)
applied
to
observations
from
eight
SEARCH
stations
in
July
2001,
and
in
January
2002
(
ESP01/
02).
Sources
of
primary
PM2.5
in
the
JST
station
are
apportioned
as
detailed
in
Table
1.
Results
from
the
two
different
methods
match
reasonably
well,
but
discrepancies
exist.
Discrepancies
can
come
from
the
direct
comparison
between
point
measurement
(
CMB­
MM)
and
the
volume­
averaged
prediction(
CMAQ),
but
more
importantly
differences
come
from
the
uncertainty
of
each
model.
Sources
of
uncertainty
in
CMB­
MM
include
source
profiles
and
ambient
data,
which
have
been
studied
by
Yan
et
al.,
(
2004:
companion
paper).
When
different
source
profiles
of
wood
burning
are
used,
contributed
masses
change
from
1.7
to
14.4ug/
m3
on
January
27,
2002.
When
different
levels
of
uncertainty
are
applied
to
the
ambient
data,
contributed
masses
from
gasoline
emissions
change
as
much
as
35%.
Sources
of
uncertainty
in
CMAQ
include
the
size
of
the
grid
emission
inventory,
and
meteorological
field
uncertainties.
As
can
be
seen
in
Table
1,
contributed
masses
are
different
up
to
43%
when
different
sizes
of
the
grids
are
applied.
Emission
inventory
uncertainties
include
not
only
the
total
mass
of
emissions,
but
the
temporal
and
compositional
profiles
and
spatial
surrogate
as
well.
At
this
point,
it
is
difficult
to
assess
which
approach
is
"
better",
and
indeed,
they
both
have
their
strengths
and
limitations,
which
are
further
explored.
2
Table
1.
Source
apportionments
of
primary
PM2.5
in
the
JST
station
using
CMB­
MM
and
CMAQ.
Unit:[
ug/
m3]

July
2001
January
2002
CMB­
MM
CMAQ(
12/
36km)
CMB­
MM
CMAQ(
12/
36km)
Diesel
exhaust
1.35
2.28/
1.48
1.84
2.17/
1.44
Gasoline
exhaust
0.07
0.61/
0.48
0.60
0.65/
0.50
Road
dust
0.44
2.29/
1.97
0.00
2.86/
2.45
Wood
burning
0.33
1.96/
2.34
3.09
2.24/
2.67
Sec.
Sulfate
7.85
10.76/
11.37
2.21
3.56/
3.19
Sec.
Nitrate
0.44
1.77/
1.38
1.65
8.05/
8.23
Sec.
Ammonium
3.26
4.36/
4.18
1.34
3.68/
3.60
Other
organic
matter
3.35
2.76/
3.23
1.47
2.55/
2.25
Unclassified
mass
5.44
3.52/
2.99
1.08
4.53/
3.74
Table
2.
Source
apportionment
of
primary
and
secondary
PM2.5
in
the
JST
station
using
CMAQ.
Unit:[
ug/
m3]

July
2001
January
2002
CMAQ(
12/
36km)
CMAQ(
12/
36km)
Diesel
exhaust
3.94/
3.51
3.88/
3.02
Gasoline
exhaust
3.45/
2.88
3.24/
3.02
Road
dust
2.46/
2.17
3.09/
2.67
Wood
burning
2.39/
2.95
2.94/
3.47
Power
plant
9.25/
10.21
3.09/
3.26
Unclassified
mass
8.83/
7.70
14.04/
12.62
